To a decent amount of fanfare, a search engine named Cuil (pronounced cool) was launched this week. Founded by ex-Googlers, Cuil was positioned as a direct competitor to Google.
Now, the web is full of articles describing the problems encountered by Cuil and it’s users in it’s first week.
While it claims to have indexed over 120 BILLION pages, it doesn’t seem to know what to do with them. Having previously worked in the search space, I know there are a few basic things that search engines need to be able to do:
1. Actually show results to common search queries
The following screenshot shows the actual results for the search terms “cuil search engine launch”
(click to enlarge)
Shockingly it came up with no results found, which makes no sense. This happened consistently, so it’s not simply a glitch due to heavy load.
2. Update your index more than once a week
Another example. Fully 5 days after the launch of Cuil, a search for the terms “cuil launch” gives only 24 hits that look like this:
(click to enlarge)
Note the complete lack of any hits related to the actual launch of the search engine itself. Not even the site itself Cuil.com shows up in the results. That tells me that while they may have indexed 120 Billion pages, they’ve indexed virtually nothing new in the past 5 days. Rather unimpressive, as relevancy and currency are both necessary for modern search engines. As a comparison, the following search engines have far more matches to the same query:
- Google: 364,000 hits
- Yahoo: 5,970,000 hits
- Live: 248,000 hits
- Altavista: 5,870,000 hits
- Excite: 56 hits (not very exciting!)
- Ask: 5460 hits
There’s a huge disparity in the numbers, but the exception of Excite — does anyone use Excite any more? — it’s pretty clear that there are hundreds of thousands, if not millions of pages on the web that match the search “cuil launch”. Even Excite’s paltry 56 hits is more than double that of Cuil.
3. Show results in a meaningful way
Cuil shows search results in a 3 column magazine style format. This is different the the standard ranked list of search results that virtually every other search engine provides. I’m not sure why Cuil’s makers decided to list their results this way. Do I assume that relevance is by column or by row? A search for “Ethan Alan Saeed” — I’ve got to assume that those three names don’t appear together that often outside of our blog — gave the following:
(click to enlarge)
So which are more or less relevant? Is Cuil saying it doesn’t matter? And why is Pop Matters listed first? Yes, that page has all three names on it, but other than that, nothing.
Google, Live and Yahoo all list our blog first.
4. Images should be relevant to the search
Most search engines show images in some of the general web hits. Cuil is no different. But what I can’t figure out is what is the relationship between hits and the associated images. Take the following search — “saeed khan product management” and this set of search results:
(click to enlarge)
Each of the hits above seems to relate to me or our blog, but careful inspection shows that not to be the case. For example, the third hit on the top links to this page on Jeff Lash’s GoodProductManager.com site.
But the actual page, from January 2007 doesn’t contain either my first name or my last name. Odd.
The other odd thing about the page is that the images associated with each link are not from those pages themselves, or even from those domains. For example, the Contact and About Us link — second row first column — has a prominent picture of someone. First of all, that image is not on that page, check for yourself here, and secondly, that picture is not a picture of me, and it is certainly not Ethan or Alan.
In fact, it is a picture of an Australian politician named Saeed Khan. No relation to me or our blog at all. As I look across the results on that page, it is clear to me that not a single one of the images has any relation to the link it is associated with.
My final thoughts
It’s hard to believe that Cuil was launched with such fanfare, yet so prematurely. One has to wonder why? Did someone or something force their hand to launch earlier than planned or was this simply an example of a really bad launch. The name is odd, and misspelling it — ciul.com or culi.com — takes you either to a parked domain or a domain that focuses on (ahem) “parking“.
But aside from that, the various problems with the site could have been alleviated by using a simple 4-letter word: BETA.
While product success certainly depends on the product, it depends even more on setting expectations and ensuring that users and customers know what to expect when they use it. There’s no good reason why Cuil couldn’t have slapped the word BETA on their site, as Google is won’t to do for new services — and thus set expectations accordingly.
Another thing they should have done is not position themselves as a direct competitor to Google. Taking on an established and dominant competitor out of the gate, like conducting a land war in Asia, is a bad strategy.
They should have followed the axiom Nail it, then Scale it.
Let’s see how they recover from this.
Saeed